Linux MIPS Architecture development
 help / color / mirror / Atom feed
From: Ralf Baechle <ralf@linux-mips.org>
To: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
Cc: David Daney <ddaney@caviumnetworks.com>,
	linux-mips@linux-mips.org,
	Peter Zijlstra <a.p.zijlstra@chello.nl>,
	Paul Mackerras <paulus@samba.org>, Ingo Molnar <mingo@elte.hu>,
	Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: Re: [PATCH v2 4/4] MIPS: perf: Add support for 64-bit perf counters.
Date: Thu, 17 Feb 2011 14:36:00 +0100	[thread overview]
Message-ID: <20110217133559.GA12732@linux-mips.org> (raw)
In-Reply-To: <AANLkTikTV-=A8H=h_F+025VB37tHSmxpsNCGndi_dAFW@mail.gmail.com>

On Thu, Feb 17, 2011 at 06:46:39PM +0800, Deng-Cheng Zhu wrote:

> The reason of the perf-record failure on 32bit platforms is that the 32bit
> counter read function mipsxx_pmu_read_counter() returns wrong 64bit values.
> For example, the counter value 0x12345678 will be returned as
> 0xffffffff12345678. So in mipspmu_event_update(), the delta will be wrong.
> So here's a possible fix for your reference:
> 
> --- a/arch/mips/kernel/perf_event_mipsxx.c
> +++ b/arch/mips/kernel/perf_event_mipsxx.c
> @@ -184,19 +184,21 @@ static unsigned int
> mipsxx_pmu_swizzle_perf_idx(unsigned int idx)
>         return idx;
>  }
> 
> +#define U32_MASK 0xffffffff
> +
>  static u64 mipsxx_pmu_read_counter(unsigned int idx)
>  {
>         idx = mipsxx_pmu_swizzle_perf_idx(idx);
> 
>         switch (idx) {
>         case 0:
> -               return read_c0_perfcntr0();
> +               return read_c0_perfcntr0() & U32_MASK;
>         case 1:
> -               return read_c0_perfcntr1();
> +               return read_c0_perfcntr1() & U32_MASK;
>         case 2:
> -               return read_c0_perfcntr2();
> +               return read_c0_perfcntr2() & U32_MASK;
>         case 3:
> -               return read_c0_perfcntr3();
> +               return read_c0_perfcntr3() & U32_MASK;

read_c0_perfctrl0 etc. are defined in mipsregs.h as 32-bit reads returning
a signed int.  That was ok on 32-bit kernels.  To support the optional
64-bit counters the code will have to be changed to something like:

static u64 mipsxx_pmu_read_counter(unsigned int idx)
{
	idx = mipsxx_pmu_swizzle_perf_idx(idx);

	switch (idx) {
	case 0:
		if (read_c0_perfctrl0() & M_PERFCTL_WIDE)
			return read_c0_64_bit_perfcntr0();
		else
			return read_c0_32_bit_perfcntr0();
	case 1:
		if (read_c0_perfctrl1() & M_PERFCTL_WIDE)
			return read_c0_64_bit_perfcntr1();
		else
			return read_c0_32_bit_perfcntr1();
...

And read_c0_32_bit_perfcntrX need to zero-extend their return value.

  Ralf

  reply	other threads:[~2011-02-17 13:36 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-01-21 22:59 [PATCH v2 0/4] MIPS: perf: Add support for 64-bit MIPS hardware counters David Daney
2011-01-21 22:59 ` [PATCH v2 1/4] MIPS: Add accessor macros for 64-bit performance counter registers David Daney
2011-01-21 22:59 ` [PATCH v2 2/4] MIPS: perf: Cleanup formatting in arch/mips/kernel/perf_event.c David Daney
2011-01-21 22:59 ` [PATCH v2 3/4] MIPS: perf: Reorganize contents of perf support files David Daney
2011-01-21 22:59 ` [PATCH v2 4/4] MIPS: perf: Add support for 64-bit perf counters David Daney
2011-01-25  3:42   ` Deng-Cheng Zhu
2011-01-26  0:20     ` David Daney
2011-01-27  6:24       ` Deng-Cheng Zhu
2011-01-27 18:41         ` David Daney
2011-01-28  2:46           ` Deng-Cheng Zhu
2011-02-17 10:46             ` Deng-Cheng Zhu
2011-02-17 13:36               ` Ralf Baechle [this message]
2011-02-17 15:26                 ` Deng-Cheng Zhu
2011-02-17 17:26               ` David Daney
2011-02-17 19:23               ` David Daney

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20110217133559.GA12732@linux-mips.org \
    --to=ralf@linux-mips.org \
    --cc=a.p.zijlstra@chello.nl \
    --cc=acme@redhat.com \
    --cc=ddaney@caviumnetworks.com \
    --cc=dengcheng.zhu@gmail.com \
    --cc=linux-mips@linux-mips.org \
    --cc=mingo@elte.hu \
    --cc=paulus@samba.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox